Medium

Vishal Rajput

Featured Book

Ultimate Neural Network Programming with Python

Ultimate Neural Network Programming with Python

Build Neural Networks from scratch and create a mathematical foundation of all the AI-related concepts

2023

Stories

Pinned

In

AIGuys

by

Vishal Rajput

·

May 7

Burning a Transformer into Silicon: The Case for GPU-Free AI Inference

A developer burned a transformer into a $50 FPGA and ran it at 53,000 tokens/sec — no GPU, no Python, no runtime. Here’s what that actually…

Burning a Transformer into Silicon: The Case for GPU-Free AI Inference

Pinned

In

AIGuys

by

Vishal Rajput

·

Apr 23, 2025

A Practical Guide To Building Agents

Guidlelines for building real world scalable Agentic AI pipelines

A Practical Guide To Building Agents

In

AIGuys

by

Vishal Rajput

·

Jul 2

D4RT (CVPR 2026 Best Paper) Changed How AI Models Understand Time, Motion, and The Physical World

Here is a question that sounds simple and isn’t. You watch a five-second video of a swan gliding across a pond. Can you say, for every…

D4RT (CVPR 2026 Best Paper) Changed How AI Models Understand Time, Motion, and The Physical World

In

AIGuys

by

Vishal Rajput

·

Jul 1

There Has Been a Situation in AI And It’s Worse Than the Hype

Anthropic shipped a model programmed to secretly make itself dumber when it catches you doing AI research. That’s not a safety feature…

There Has Been a Situation in AI And It’s Worse Than the Hype

In

AIGuys

by

Vishal Rajput

·

Jun 22

Claude, GPT & Gemini Are Loosing: Intelligence Is Getting Commoditized

OpenSource AI models are delivering 90% of the performance of top U.S. AI models while costing only 1/5th.

Claude, GPT & Gemini Are Loosing: Intelligence Is Getting Commoditized

In

AIGuys

by

Vishal Rajput

·

Jun 19

NVIDIA Proved 4-Bit Training Works at Real Scale (Not Just Inference)

A 12-billion-parameter Mamba-Transformer, trained on 10 trillion tokens, entirely in 4-bit floating point, matched an FP8 baseline almost…

NVIDIA Proved 4-Bit Training Works at Real Scale (Not Just Inference)

In

AIGuys

by

Vishal Rajput

·

Jun 16

Why No One Talks About LangChain, LangGraph Or AutoAgent

The post-framework consensus arrived faster than I expected, and it arrived smarter than the early industry predictions. Most production…

Why No One Talks About LangChain, LangGraph Or AutoAgent

In

AIGuys

by

Vishal Rajput

·

Jun 10

One Model, Two Products: Fable 5 and Mythos From Anthropic

Anthropic shipped Fable 5 and Mythos 5 today. They’re the same model. Who gets which version depends entirely on who you are. That’s a…

One Model, Two Products: Fable 5 and Mythos From Anthropic

In

AIGuys

by

Vishal Rajput

·

Jun 10

Google Just Shrunk 31 GB of AI Memory to 4 GB. Here’s the Math.

TurboVec is an open-source vector index built on Google Research’s TurboQuant algorithm: 16× compression, faster than FAISS, zero training…

Google Just Shrunk 31 GB of AI Memory to 4 GB. Here’s the Math.

In

AIGuys

by

Vishal Rajput

·

Jun 8

We Are Getting So Much Wrong With Current AI

I’ve been writing and building AI pipelines and models since the launch of TensorFlow 1. But lately it has started feeling a bit dull. Am I…

We Are Getting So Much Wrong With Current AI

Vishal Rajput

Vishal Rajput

Book Author

3x🏆Top writer in AI | AI Book 📓: https://rb.gy/xc8m46 | LinkedIn +: https://www.linkedin.com/in/vishal-rajput-999164122/ | 𝕏: https://x.com/RealAIGuys

Help

Status

About

Careers

Press

Blog

Store

Privacy

Rules

Terms

Text to speech